Remote monitoring method for working state of coal mining machine

Abstract

The invention discloses a remote monitoring method for the working state of a coal mining machine, and the method comprises the steps: constructing an SDG model of fault evolution of the coal mining machine, constructing a multi-dimensional causal relationship model through the SDG model, and analyzing the relevance between different monitoring parameters and faults; historical operation data of the coal mining machine are collected, and a semi-supervised learning model is constructed; historical multi-source data of the coal mining machine are collected, a digital twinborn body of the coal mining machine is constructed, and the digital twinborn body and a mechanism model of the coal mining machine are fused; and collecting real-time operation data of the coal mining machine, and respectively inputting the real-time operation data into the multi-dimensional causal relationship model, the semi-supervised learning model and the digital twinborn body of the coal mining machine for monitoring to obtain a monitoring result. According to the invention, errors caused by manual judgment are reduced, and scientificity and effectiveness of decision making are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of coal mine mechanical equipment monitoring, and in particular to a method for remotely monitoring the working state of a coal mining machine. Background Art

[0002] As the core equipment of coal mining face, the operation status of coal mining machine is directly related to the production efficiency and safety of coal mining. However, due to the harsh working environment, many parts and complex structure of coal mining machine, traditional monitoring methods often fail to timely and accurately grasp its working status, resulting in frequent failures and affecting production.

[0003] In the prior art, although there are systems for remote monitoring of coal mining machines, there are still deficiencies in fault prediction, monitoring parameter optimization, and model training under low-label data. The limitations of traditional monitoring methods are: 1. Single monitoring parameters: Traditional monitoring methods often only focus on a few key parameters, such as current and temperature, while ignoring other parameters that may affect the operating status of coal mining machines. This single parameter monitoring method is difficult to fully reflect the actual operating status of coal mining machines. 2. Insufficient data processing capabilities: Traditional monitoring systems have limitations in data processing and cannot quickly and accurately analyze and process large amounts of real-time data. This results in the system's weak ability in fault prediction and diagnosis, and it is difficult to discover and solve potential problems in a timely manner. 3. Poor model generalization performance: In the field of machine learning and artificial intelligence, the generalization performance of the model is an important indicator to measure its performance in practical applications. However, traditional monitoring methods often lack effective model training and optimization methods, resulting in poor generalization performance of the model under low-label data or new scenarios. 4. Lack of multi-source data fusion: The operating status of coal mining machines is affected by many factors, including the status of the equipment itself, working environment parameters, and external interference. Traditional monitoring methods often only focus on a single data source and lack the ability to fuse multi-source data, making it difficult to achieve comprehensive perception and accurate judgment of the operating status of coal mining machines.

[0004] Modern monitoring technology achieves comprehensive perception of the operating status of the coal mining machine by introducing more monitoring parameters, such as vibration, sound, and images. These multi-dimensional monitoring parameters can more accurately reflect the actual operating status of the coal mining machine and provide richer information for fault prediction and diagnosis. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a remote monitoring method for the working status of a coal mining machine. Through a symbolic directed graph, the evolution process of coal mining machine faults is qualitatively analyzed, a multidimensional causal relationship model is constructed, and a semi-supervised learning model is introduced to achieve comprehensive, real-time and accurate monitoring of the working status of the coal mining machine.

[0006] To achieve the above object, the present invention provides a method for remotely monitoring the working status of a coal mining machine, comprising:

[0007] Constructing an SDG model of coal mining machine failure evolution, and constructing a multi-dimensional causal relationship model through the SDG model;

[0008] Collect historical operation data of coal mining machines and build a semi-supervised learning model;

[0009] Collect historical multi-source data of the coal mining machine, build a digital twin of the coal mining machine, and fuse the digital twin with the mechanism model of the coal mining machine;

[0010] The real-time operation data of the coal mining machine is collected, and the real-time operation data is respectively input into the multidimensional causal relationship model, the semi-supervised learning model and the digital twin of the coal mining machine for monitoring to obtain the working status of the coal mining machine.

[0011] Preferably, constructing the multidimensional causal relationship model includes:

[0012] According to the working principle of the coal mining machine, common fault types and monitoring targets, the key parameters that need to be monitored are determined, and the conditional entropy is used to measure the index of the monitoring parameters' sensitivity to faults. At the same time, the importance of parameter attributes is considered to evaluate the monitoring parameters' ability to distinguish coupled faults.

[0013] Combining conditional entropy and parameter attribute importance, a mathematical model is established to sort and screen monitoring parameters to form a functional fault model, and the SDG model of coal mining machine fault evolution is constructed using the symbolic directed graph theory;

[0014] Based on the SDG model, the correlation between different monitoring parameters and faults is analyzed, the role and weight of the monitoring parameters in fault prediction are determined, and the multidimensional causal relationship model is constructed.

[0015] Preferably, the process of constructing the SDG model also includes classifying common failure modes of coal mining machines and defining specific failure evolution paths and failure propagation directions for each type of failure mode, so as to enhance the accuracy and timeliness of the model's failure prediction.

[0016] Preferably, the process of constructing the multidimensional causal relationship model also includes environmental factors, operating habits and maintenance record influencing factors.

[0017] Preferably, the method for constructing the semi-supervised learning model is:

[0018] The historical operation data of the coal mining machine is divided into a training set and a test set, an ensemble learning method is used, pseudo labels are generated through iterative training, and based on a set confidence threshold, samples with high confidence in the prediction results are screened, and the generated pseudo labels are added to the training set as true labels, and the model is retrained using the updated training set, and the above steps are repeated until a predetermined number of iterations is met, and finally the trained model is evaluated through the test set to obtain the semi-supervised learning model;

[0019] The training set is labeled data with the operating status of the coal mining machine marked, and the test set is unlabeled data with the operating status of the coal mining machine not marked.

[0020] Preferably, an anomaly detection module is also integrated into the semi-supervised learning model, and the anomaly detection module is used to automatically discover data points that deviate from normal patterns in real-time data as potential fault warning signals.

[0021] Preferably, the dynamic behavior of the coal mining machine is simulated using a physical engine and simulation technology, and the actual operation data of the coal mining machine is synchronized in real time through the Internet of Things technology to construct a digital twin of the coal mining machine. The fusion process of the digital twin and the coal mining machine mechanism model includes parameter calibration, model verification and error compensation.

[0022] Preferably, the working status of the coal mining machine is obtained through a human-computer interaction interface, and the human-computer interaction interface is used for remote monitoring personnel to view the monitoring data, fault warning information, maintenance suggestions and historical records of the coal mining machine in real time, so as to realize transparency and visualization of information.

[0023] Preferably, the monitoring method can also perform adaptive learning to automatically optimize the parameters and structures of the multidimensional causal relationship model, the semi-supervised learning model and the digital twin to adapt to changes in the performance of the coal mining machine and the emergence of new failure modes.

[0024] Compared with the prior art, the present invention has the following advantages and technical effects:

[0025] The present invention can more accurately predict the occurrence of faults, reduce false positives and false negatives, and improve the accuracy and timeliness of predictions by classifying common fault modes and defining specific fault evolution paths and propagation directions during the SDG model construction process. The multidimensional causal relationship model not only considers the direct causes and consequences of the faults, but also incorporates multidimensional data such as environmental factors, operating habits, and maintenance records, thus achieving a comprehensive evaluation of the working status of the coal mining machine.

[0026] Based on the comprehensive output of multidimensional causal model, semi-supervised learning model and digital twin, the present invention can comprehensively evaluate the working status of the coal mining machine and provide accurate maintenance suggestions and fault handling solutions, which helps to reduce errors in human judgment and improve the scientificity and effectiveness of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0028] Figure 1 The present invention is a flowchart of a method for remotely monitoring the working status of a coal mining machine according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0030] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0031] The present invention proposes a method for remotely monitoring the working status of a coal mining machine. Figure 1 ,include:

[0032] Constructing an SDG model of coal mining machine failure evolution, and constructing a multi-dimensional causal relationship model through the SDG model;

[0033] Collect historical operation data of coal mining machines and build a semi-supervised learning model;

[0034] Collect historical multi-source data of the coal mining machine, build a digital twin of the coal mining machine, and fuse the digital twin with the mechanism model of the coal mining machine;

[0035] The real-time operation data of the coal mining machine is collected, and the real-time operation data is respectively input into the multidimensional causal relationship model, the semi-supervised learning model and the digital twin of the coal mining machine for monitoring to obtain the working status of the coal mining machine.

[0036] Furthermore, a multidimensional causal relationship model is constructed, including:

[0037] According to the working principle of the coal mining machine, common fault types and monitoring targets, the key parameters that need to be monitored are determined, and the conditional entropy is used to measure the index of the monitoring parameters' sensitivity to faults. At the same time, the importance of parameter attributes is considered to evaluate the monitoring parameters' ability to distinguish coupled faults.

[0038] Combining conditional entropy and parameter attribute importance, a mathematical model is established to sort and screen monitoring parameters to form a functional fault model, and the SDG model of coal mining machine fault evolution is constructed using the symbolic directed graph theory;

[0039] Based on the SDG model, the correlation between different monitoring parameters and faults is analyzed, the role and weight of the monitoring parameters in fault prediction are determined, and the multidimensional causal relationship model is constructed.

[0040] When constructing the SDG model, it is first necessary to comprehensively sort out and classify the common failure modes of coal mining machines. This includes but is not limited to mechanical failures (such as bearing damage, gear wear), electrical failures (such as motor overheating, line short circuit), hydraulic failures (such as abnormal oil pressure, leakage) and control system failures (such as sensor failure, program error), etc. For each type of fault, it is further subdivided into specific fault subclasses to more accurately describe the fault characteristics and evolution process. For each type (or subclass) of fault mode, a clear fault evolution path is defined. These paths should be based on the physical mechanism of the fault and the interaction relationship between the systems, and describe in detail the evolution of the fault from the initial state to the final state. In this embodiment, taking mechanical failure as an example, bearing damage may start with poor lubrication, gradually develop into temperature increase, increased vibration, and eventually lead to bearing rupture. Each evolution path contains a series of key nodes and state transition conditions to accurately represent it in the SDG model. In the SDG model, in addition to the fault evolution path, it also includes a clear identification of the propagation direction of the fault, specifically the propagation path of the fault between the components within the system, and the direction of the impact of the fault on the overall performance of the system. By introducing directed edges and signed edges (indicating positive or negative impacts), we can clearly show how faults propagate from one component to another and the positive or negative impact these faults have on system performance.

[0041] The SDG model is continuously updated and optimized as the coal mining machine operation data accumulates and the failure mode evolves, including adjusting the fault classification and evolution path according to new fault cases, updating the quantitative value of the fault influencing factor according to real-time data, and automatically identifying and predicting potential failure modes through machine learning algorithms. Through continuous updates and optimizations, the SDG model is ensured to always maintain high-precision monitoring and prediction capabilities for the working status of the coal mining machine.

[0042] In the process of constructing the multidimensional causal model, not only the direct causes and consequences of the failure are considered, but also multi-dimensional data such as environmental factors, operating habits, and maintenance records are incorporated to comprehensively evaluate the working status and potential risks of the coal mining machine.

[0043] Further, environmental factors include:

[0044] Physical environment: natural factors such as temperature, humidity, dust concentration, air pressure, etc. in the working environment of the coal mining machine, which may directly affect the mechanical and electrical properties of the coal mining machine.

[0045] Geological conditions: geological characteristics of the coal seam such as hardness, thickness, inclination, etc. These factors will affect the cutting efficiency and stability of the coal mining machine.

[0046] Operating habits include:

[0047] Operator behavior: Monitor the operator's skill level, operating habits (such as the number of emergency stops, acceleration and deceleration frequency), fatigue level, etc. These factors may affect the smoothness and safety of the coal mining machine's operation.

[0048] Human-computer interaction: Analyze the interaction data between the operator and the coal mining machine, such as the response time of operating instructions, feedback accuracy, etc., to evaluate the system's response speed and operational convenience.

[0049] Maintenance records include:

[0050] Historical failures: record the types of failures, occurrence times, treatment measures and results of the coal mining machine in history to provide a reference for predicting future failures.

[0051] Maintenance cycle: Track the regular maintenance and overhaul records of coal mining machines, including replacement of parts, adjustment of parameters, etc., to evaluate the maintenance status and service life of the equipment.

[0052] Spare parts management: Consider factors such as spare parts inventory, replacement frequency, and supplier reliability to ensure that failed parts can be replaced in a timely manner when needed.

[0053] In the process of constructing a multidimensional causal relationship model, the collected multidimensional data is preprocessed by cleaning, denoising, normalizing and other preprocessing work to improve data quality and consistency. The causal relationship chain between the data of each dimension is identified by means of statistical methods, machine learning algorithms or expert knowledge base. This includes direct causal relationships (such as motor overheating due to excessive temperature) and indirect causal relationships (such as accelerated equipment wear due to improper operation). Finally, based on the identified causal relationship chain, a multidimensional causal relationship model is constructed. The model can comprehensively consider the influencing factors of multiple dimensions and quantify their contribution to the working state of the coal mining machine. At the same time, the model should have the ability to dynamically adjust to adapt to the continuous changes in the working environment of the coal mining machine. It can realize a comprehensive assessment of the working state of the coal mining machine and timely early warning of potential risks. This not only improves the accuracy and timeliness of the monitoring system, but also provides a strong guarantee for the safe and stable operation of the coal mining machine.

[0054] Furthermore, the process of building a semi-supervised learning model is as follows:

[0055] Collect the operation data of the coal mining machine, including data in normal and abnormal states. This data may come from multiple channels such as sensors, control systems, maintenance records, etc. The data is divided into labeled data and unlabeled data. Labeled data refers to data that has been clearly marked with the operating status of the coal mining machine (normal or abnormal), and unlabeled data refers to unlabeled data.

[0056] Clean the collected data to remove noise, outliers and redundant information. Normalize the data so that data of different dimensions can be compared and analyzed on the same scale. Convert the raw data into a form that the model can process.

[0057] Train an initial ensemble learning model using a small amount of labeled data. Evaluate the performance of the model to ensure that it has a certain degree of recognition ability on labeled data.

[0058] Use the initial model to make predictions on unlabeled data and generate pseudo labels.

[0059] According to the set confidence threshold, samples with higher confidence in the prediction results are screened out, and their pseudo labels are added to the training set as true labels.

[0060] The model is retrained using the updated training set to improve the model's ability to identify abnormal operating conditions of the coal mining machine.

[0061] Repeat the above steps until the predetermined number of iterations is met or the model performance reaches a stable state.

[0062] This embodiment also includes adopting a self-training strategy, that is, first using labeled data to train an initial model, then using the model to predict unlabeled data, adding samples with higher confidence in the prediction results as pseudo labels to the training set, and iteratively training the model.

[0063] The optimized model is deployed to the remote monitoring system of the coal mining machine's working status to monitor the operating status of the coal mining machine in real time.

[0064] Furthermore, an anomaly detection module is integrated into the semi-supervised learning model. The anomaly detection module is used to automatically discover data points that deviate from the normal pattern in real-time data as early warning signals of potential faults.

[0065] Specifically, before real-time data is input into the anomaly detection module, data preprocessing is required, including filtering, denoising, feature extraction, etc., to reduce the impact of noise and redundant information on anomaly detection.

[0066] The anomaly detection module is trained using historical data (including labeled and unlabeled data). First, the labeled data is used for supervised pre-training, and then the unlabeled data is used for unsupervised fine-tuning to obtain a trained anomaly detection module.

[0067] The pre-processed real-time data is input into the trained anomaly detection module to automatically calculate the anomaly score or probability. It determines whether the data point is abnormal based on the set threshold and generates a corresponding warning signal.

[0068] When an anomaly is detected, the early warning mechanism is triggered immediately, such as sending alarm information to relevant personnel, recording abnormal events, adjusting monitoring strategies, etc. At the same time, the severity of the fault and the possible scope of impact can be evaluated based on the type of anomaly and historical data, providing guidance for subsequent troubleshooting and maintenance. As the working status of the coal mining machine changes and new data accumulates, the training data and model parameters of the anomaly detection module need to be updated regularly to maintain its adaptability and accuracy. At the same time, the early warning signals need to be continuously monitored and analyzed to discover potential room for improvement and new failure modes.

[0069] Collect historical multi-source data of the coal mining machine, build a digital twin of the coal mining machine, and integrate the digital twin with the mechanism model of the coal mining machine, including:

[0070] First, use advanced physics engines (such as Unity's PhysX, Unreal Engine's Havok, etc.) to build a 3D physical model of the coal mining machine. The 3D physical model must accurately reflect the mechanical structure, mass distribution, material properties, and dynamic characteristics of the coal mining machine to ensure the accuracy and authenticity of the simulation results.

[0071] The dynamic behavior simulation of the coal mining machine under different working conditions is realized through programming, including but not limited to key actions such as walking, cutting, loading, and transportation. The impact of environmental factors (such as coal seam hardness, inclination, and humidity) on the performance of the coal mining machine is considered during the simulation process, as well as the interaction between mechanical components, such as gear meshing and bearing friction.

[0072] Combining historical fault data and expert knowledge base, possible fault scenarios such as motor overheating and transmission system failure are simulated in a simulation environment to provide a basis for formulating countermeasures in advance.

[0073] Various sensors (such as temperature sensors, vibration sensors, pressure sensors, position sensors, etc.) are widely deployed on coal mining machines to monitor various operating parameters and status information of coal mining machines in real time. The collected data is transmitted to the remote monitoring center in real time through Internet of Things technologies (such as LoRa, NB-IoT, 5G, etc.) to ensure the timeliness and integrity of the data. At the data receiving end, the data from different sensors are fused and processed to eliminate redundant and contradictory information, and data verification is performed at the same time to ensure the accuracy and reliability of the data.

[0074] Furthermore, the real-time data transmitted by the Internet of Things is mapped to the digital twin of the coal mining machine to achieve synchronous updates of the virtual and real worlds. The monitoring personnel can intuitively see the current status and future trends of the coal mining machine through the digital twin. The real-time data is analyzed and processed using technologies such as big data analysis and machine learning to identify abnormal patterns and potential risks and issue early warning signals. At the same time, the operation strategy of the coal mining machine is optimized based on the analysis results to improve production efficiency and safety. In an emergency, the monitoring personnel can directly intervene in the operation of the coal mining machine through the remote control system, such as adjusting the cutting speed, changing the walking path, etc., to ensure safe and smooth production.

[0075] The fusion process of the digital twin and the coal mining machine mechanism model includes parameter calibration, model verification and error compensation, specifically:

[0076] Parameter calibration: Collect key operating parameters from the shearer’s sensor network, including motor current, temperature, vibration frequency, cutting force, etc. Map the collected actual parameters to the corresponding parameters in the digital twin to establish a clear correspondence. Select the least squares method to iteratively adjust the parameters in the digital twin until the error between it and the actual parameters reaches an acceptable range. As the shearer continues to operate, its parameters may change. Therefore, a real-time update mechanism is established to ensure that the parameters in the digital twin can be updated synchronously with the changes in the actual parameters.

[0077] Model verification: Under the same or similar working conditions, record the actual operation data of the coal mining machine and the simulation data of the digital twin at the same time. Compare and analyze the two sets of data, calculate the relative error or absolute error of each parameter, and evaluate the accuracy of the digital twin. Analyze the influence of different parameters on the simulation results, and determine which parameters are more sensitive to the accuracy of the model so as to focus on them in subsequent optimization. Write a verification report based on the verification results, record the verification process, problems found, and improvement suggestions. Verify the accuracy and reliability of the digital twin in the simulation of the operation of the coal mining machine.

[0078] Error compensation: Identify the source of errors in the digital twin through comparative testing or sensitivity analysis. Design corresponding compensation strategies based on the source of errors. In this embodiment, data fusion or filtering technology can be used to process errors caused by insufficient sensor accuracy; for errors caused by model simplification, a more complex physical model can be introduced for correction. Apply the compensation strategy to the digital twin, re-run the simulation test, and verify the compensation effect. Based on feedback from actual applications and new research results, continuously optimize the error compensation strategy and digital twin model to ensure that it can always accurately reflect the actual operating status and performance changes of the coal mining machine.

[0079] Through the close combination of parameter calibration, model verification and error compensation, the deep integration of the digital twin and the coal mining machine mechanism model can be ensured, providing an accurate and reliable basis for remote monitoring.

[0080] The causal relationship analysis in the multidimensional causal relationship model, the intelligent prediction in the semi-supervised learning model, and the performance evaluation and fault warning in the digital twin are combined to comprehensively evaluate the working status of the coal mining machine.

[0081] According to the evaluation results, intelligent algorithms (such as decision trees, support vector machines, deep learning, etc.) are used to make intelligent decisions and give the best maintenance suggestions or fault handling solutions. A friendly human-computer interaction interface is provided so that monitoring personnel can intuitively understand the evaluation results and decision suggestions, and make adjustments or confirmations according to actual conditions.

[0082] The monitoring method can also perform adaptive learning to automatically optimize the parameters and structures of the multidimensional causal relationship model, semi-supervised learning model, and digital twin to adapt to changes in coal mining machine performance and the emergence of new failure modes. Adaptive learning not only optimizes individual models, but also promotes the synergy between multidimensional causal relationship models, semi-supervised learning models, and digital twins. Through cross-model information sharing and complementarity, the overall performance is optimized.

[0083] The present invention can fully utilize the advantages of multidimensional causal relationship models, semi-supervised learning models and digital twins to provide comprehensive, accurate and intelligent decision support for remote monitoring of the working status of coal mining machines.

[0084] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for remotely monitoring the working status of a coal mining machine, characterized in that: include: Constructing an SDG model of coal mining machine failure evolution, and constructing a multi-dimensional causal relationship model through the SDG model; Collect historical operation data of coal mining machines and build a semi-supervised learning model; Collect historical multi-source data of the coal mining machine, build a digital twin of the coal mining machine, and fuse the digital twin with the mechanism model of the coal mining machine; The real-time operation data of the coal mining machine is collected, and the real-time operation data is respectively input into the multidimensional causal relationship model, the semi-supervised learning model and the digital twin of the coal mining machine for monitoring to obtain the working status of the coal mining machine.

2. The method for remotely monitoring the working status of a coal mining machine according to claim 1, characterized in that: Constructing the multidimensional causal relationship model includes: According to the working principle of the coal mining machine, common fault types and monitoring targets, the key parameters that need to be monitored are determined, and the conditional entropy is used to measure the index of the monitoring parameters' sensitivity to faults. At the same time, the importance of parameter attributes is considered to evaluate the monitoring parameters' ability to distinguish coupled faults. Combining conditional entropy and parameter attribute importance, a mathematical model is established to sort and screen monitoring parameters to form a functional fault model, and the SDG model of coal mining machine fault evolution is constructed using the symbolic directed graph theory; Based on the SDG model, the correlation between different monitoring parameters and faults is analyzed, the role and weight of the monitoring parameters in fault prediction are determined, and the multidimensional causal relationship model is constructed.

3. The method for remotely monitoring the working status of a coal mining machine according to claim 2, characterized in that: The process of constructing the SDG model also includes classifying common failure modes of coal mining machines and defining specific failure evolution paths and failure propagation directions for each type of failure mode, thereby enhancing the accuracy and timeliness of the model's failure prediction.

4. The method for remotely monitoring the working status of a coal mining machine according to claim 3, characterized in that: The process of constructing the multidimensional causal relationship model also includes environmental factors, operating habits and maintenance record influencing factors.

5. The method for remotely monitoring the working status of a coal mining machine according to claim 1, characterized in that: The method for constructing the semi-supervised learning model is: The historical operation data of the coal mining machine is divided into a training set and a test set, an ensemble learning method is used, pseudo labels are generated through iterative training, and based on a set confidence threshold, samples with high confidence in the prediction results are screened, and the generated pseudo labels are added to the training set as true labels, and the model is retrained using the updated training set, and the above steps are repeated until a predetermined number of iterations is met, and finally the trained model is evaluated through the test set to obtain the semi-supervised learning model; The training set is labeled data with the operating status of the coal mining machine marked, and the test set is unlabeled data with the operating status of the coal mining machine not marked.

6. The method for remotely monitoring the working status of a coal mining machine according to claim 5, characterized in that: An anomaly detection module is also integrated into the semi-supervised learning model, and the anomaly detection module is used to automatically find data points that deviate from normal patterns in real-time data as potential fault warning signals.

7. The method for remotely monitoring the working status of a coal mining machine according to claim 1, characterized in that: The dynamic behavior of the coal mining machine is simulated by using a physical engine and simulation technology, and the actual operation data of the coal mining machine is synchronized in real time through the Internet of Things technology to construct a digital twin of the coal mining machine. The fusion process of the digital twin and the coal mining machine mechanism model includes parameter calibration, model verification and error compensation.

8. The method for remotely monitoring the working status of a coal mining machine according to claim 1, characterized in that: The working status of the coal mining machine is obtained through a human-computer interaction interface, and the human-computer interaction interface is used for remote monitoring personnel to view the monitoring data, fault warning information, maintenance suggestions and historical records of the coal mining machine in real time, so as to realize transparency and visualization of information.

9. The method for remotely monitoring the working status of a coal mining machine according to any one of claims 1 to 8, characterized in that: The monitoring method can also perform adaptive learning and automatically optimize the parameters and structures of the multidimensional causal relationship model, the semi-supervised learning model, and the digital twin to adapt to changes in the performance of the coal mining machine and the emergence of new failure modes.

Images